Social Science & Medicine 56 (2003) 785–792

Does active leisure protect cognition? Evidence from a nationalbirth cohort

Marcus Richards*, Rebecca Hardy, Michael E.J. Wadsworth

Department of Epidemiology and Public Health, University College London, MRC National Survey of Health and Development,

1-19 Torrington Place, London WC1E 6BT, UK

Abstract

Social, physical and intellectual activities are thought to facilitate cognitive performance and slow the rate of age

associated cognitive decline, but little is known about this association in younger adulthood. We used multiple

regression to test the association between two kinds of activity at 36 years—physical exercise and spare-time activity—

and verbal memory at 43 and 53 years in 1919 males and females enrolled in the MRC National Survey of Health and

Development (the British 1946 birth cohort). Both kinds of activities were significantly and positively associated with

memory performance at 43 years, after controlling for sex, education, occupational social class, IQ at 15 years, and

recurrent ill health and significant mental distress. Furthermore, physical exercise at 36 years (but not spare-time

activity) was associated with a significantly slower rate of decline in memory from 43 to 53 years, after controlling for

the same factors, with evidence that continuing physical exercise after 36 years was important for protection. We

conclude that physical exercise and spare-time activity are significantly associated with benefit to memory in midlife,

although these two kinds of voluntary activity may exert their effects on cognition via different paths.r 2002 Elsevier

Science Ltd. All rights reserved.

Keywords: Spare-time activity; Physical exercise; Verbal memory; Birth cohort; Britain

Introduction

The identification of controllable factors that modify

rate of age-associated cognitive decline has major public

health implications. The suggestion that engagement in

challenging spare-time activities can diminish (Schaie,

1984), and perhaps even reverse (Schaie & Willis, 1986)

the rate of this decline is therefore of considerable

importance. There are, however, several theoretical and

methodological problems in interpreting findings in this

area.

First, it is unclear which activities are most important

for cognitive maintenance. Significant effects have been

reported for intellectual stimulation (Arbuckle, Gold,

Andres, Schwartzman, & Chaikelson, 1992; Hultsch,

Hertzog, Small, & Dixon, 1999), social engagement

(Bassuk, Glass, & Berkman, 1999) and physical exercise

(Albert et al., 1995; Carmelli, Swan, LaRue, & Eslinger,

1997; Kramer et al., 1999). Unfortunately these different

components are not easy to isolate from each other. This

is important, however, for understanding biological

pathways to neural development, and for formulating

possible intervention strategies for minimising cognitive

decline.

Second, direction of causality is problematic. While it

is possible that activity directly promotes cognitive

growth and maintenance, those with high cognitive

ability are also likely to engage in such activities.

Similarly, decline in cognitive performance may lead to

diminution of activity. This issue is difficult to resolve,

even with longitudinal data (Hultsch et al., 1999).

Third, some studies do not control for factors

that may account for the association between

activity and cognition, particularly education and

occupational social class (Schaie, 1984; Gold et al.,

1995).

*Corresponding author. Tel.: +44-207-679-1737; fax: +44-

207-813-0280.

E-mail address: m.richards@ucl.ac.uk (M. Richards).

0277-9536/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.

PII: S 0 2 7 7 - 9 5 3 6 ( 0 2 ) 0 0 0 7 5 - 8

Fourth, it is unclear whether there is a sensitive period

for activity engagement. Most studies have focused on

later life, and little is known about the effects of activity

on cognition in earlier adulthood. It is particularly

important to determine if patterns of activity established

early in life are more effective than activities begun later,

and whether any benefit of this activity depends on the

extent to which it is sustained.

The Medical Research Council National Survey of

Health and Development (the British 1946 birth cohort)

provides an opportunity to examine effects of activity on

memory performance in a pre-retirement population.

Participants were asked about physical exercise and

spare-time activities at 36 years. Verbal memory was

then measured at 43 and 53 years. We investigated the

association between the above two types of activity and

verbal memory at these ages, controlling for sex,

educational attainment, socioeconomic status, and

mental and physical health status. In addition, general

intellectual ability was measured at 15 years in this

cohort. Since this is highly correlated with adult IQ

(Birren, Kinney, Schaie, & Woodruff, 1981; Deary,

Whalley, Lemmon, Crawford, & Starr, 2000) we there-

fore had a rare opportunity to adjust these analyses for

baseline IQ. This would reduce the likelihood that any

apparent effect of activity on memory arose from those

of high cognitive ability simply being more likely to

engage in stimulating activities.

Method

Participants

Participants comprised the study population of the

MRC National Survey of Health and Development

(NSHD), also known as the British 1946 birth cohort,

and initially consisting of 5362 children of non-manual

and agricultural workers and a random sample of one in

four of manual workers selected from all single and

legitimate births that occurred in England, Scotland and

Wales during one week in March 1946 (Wadsworth,

1991). The cohort has been studied on 21 occasions

between birth and 53 years, with information about

sociodemographic factors and medical, cognitive and

psychological function being obtained by interview,

examination and postal questionnaire. The most recent

contact was an interview in 1999 at 53 years, when

sample size was 3035. In 1989 the cohort was shown to

be a representative sample, in most respects, of the UK

population legitimately and singly born in the immediate

post-war era (Wadsworth et al., 1992). Exceptions were

an over-representation among non-responders of the

never married, the least literate, those always in manual

social class circumstances, and those with psychiatric

illness (Wadsworth et al., 1992). By 43 years permanent

losses comprised 365 (6.8%) deaths, and 540 (10.1%)

refusals, and temporary losses were 607 (11.3%)

emigrations or residence overseas and 370 (6.9%)

failures to contact. Losses through death have been

greater among those in manual rather than non-manual

social classes (Pless, Crips, Davies, & Wadsworth, 1989).

Data collection

All measures in this study were obtained by research

nurses, who underwent standardised training for all

measures, including cognitive testing.

Measures

Outcome: verbal memory at 43 and 53 years

Cognitive function was assessed at 43 and 53 years by

measures of memory, speed and concentration. Memory

was chosen as the outcome measure because prominence

is assigned to memory in classification criteria for mild

cognitive impairment (Ritchie & Touchon, 2000) and

dementia (American Psychiatric Association, 1994), and

because poor memory, unlike letter search and praxis, is

a predictor of clinically significant cognitive decline

(Masur, Sliwinski, Lipton, Blau, & Crystal, 1994; Jacobs

et al., 1995). At both ages this was assessed by a 15-item

word learning task devised by the NSHD. Each word

was shown for 2 s. When all 15 words were shown, the

cohort member was asked to write down as many of the

words as possible. This was repeated twice more. Total

number of words correctly recalled over the three trials

(max=45) was summed to provide an overall score for

this test.

Exposure: activities at 36 years

(1) Physical exercise: Questions about physical

exercise were based on the Minnesota leisure time

physical activity questionnaire (Taylor et al., 1978).

These asked about engagement in sports and recrea-

tional activities in the previous month, utilising a

checklist of 25 activities (Table 1). Responses were

coded into no activity, 1–4 activities or 5+ activities per

month (Kuh & Cooper, 1992). Preliminary analysis

failed to demonstrate a dose–response effect with

memory across these three categories, resulting in this

measure being dichotomised to none versus any.

A similar measure was available at 43 years, although

this was based on answers to an open-ended question

about sports, vigorous leisure activities or exercises,

rather than the above checklist. However, responses

were able to be categorised in a similar way to the 36

year measure.

(2) Spare-time activity. Participants were asked about

current engagement (yes/no) in seven spare time

activities (Table 1). A total spare-time activity score

was obtained by summing these items. Because of

M. Richards et al. / Social Science & Medicine 56 (2003) 785–792786

diminishing numbers engaging in more than three

activities, this score was recoded into 0, 1, 2 and 3+

activities. Preliminary analysis showed that the largest

increase in memory was associated with a change from 0

to 1 activity, with relatively small increases across the

remaining two categories. The score was therefore

dichotomised to none versus any.

Spare-time activities were also enquired about at 43

years. However, (a) and (c) (Table 1) were omitted, an

item about constructive activities (e.g. making things by

hand) was added, and (f) (Table 1) was broadened to

include musical, artistic or other creative activities (not

necessarily with others). Post-hoc analysis (see Results)

revealed that the resulting spare-time activity measure

was not comparable to the one at 36 years.

Potential confounding variables

1. Sex, and educational and occupational attainment:

educational attainment by 26 years was classified as

no qualification, below ordinary secondary qualifica-

tions, ordinary secondary qualifications (‘O’ levels

and their training equivalents), advanced secondary

education (‘A’ levels and their equivalents), or higher

education (degree level or equivalent). Occupational

social class at 36 years, or between 26 and 36 years if

this was unknown, was classified according to the

Registrar General, and dichotomised into manual vs.

non-manual.

2. General intellectual ability: No measure of IQ was

available at 36 years. However, tests of verbal and

non-verbal ability were given at 8, 11 and 15 years.

Of these, the most comprehensive measures were

given at 15 years (Pidgeon, 1968). At this age, verbal

and non-verbal intelligence (AH4) was assessed,

along with reading comprehension (Watts–Vernon),

and mathematics. Summing these test scores derived

a global score, representing general intellectual

ability.

3. Health status at 36 years: At 36 years participants

were asked if they had any recurrent physical

problems (i.e. ‘all or most of the time’), using a

checklist of 21 items (Table 2). A total score was

obtained and recoded into 0, 1 and 2+ problems,

which yielded approximately equal numbers in each

category. Mental distress was measured by the

Present State Examination (Wing, Cooper, & Sartor-

ius, 1974), and participants were classified into cases

or non-cases according to the index of definition

(Wing & Sturt, 1978), using the cut-off for threshold

disorder or above. The potential confounding effects

of employment status, marital status, number of

children living in the home, body mass index, resting

pulse rate, systolic and diastolic blood pressure or

peak expiratory flow rate (PEFR), all at 36 years,

were also investigated.

4. Cardiovascular risk at 53 years: At 53 years partici-

pants were classified with, or at risk for, cardiovas-

cular disease if any of the following self-reported

conditions were present over the previous 10 years:

angina, coronary thrombosis, myocardial infarction,

valvular disease, aortic stenosis, ischaemic heart

disease, tachycardia, palpitations or heart murmur,

Table 1

Physical exercise and spare-time activities at 36 years

Physical exercise

Badminton Table tennis

Bowls Tennis

Cricket Yoga

Exercises like press ups,

sit ups (gym or home)

Water skiing

Football (including

refereeing)

Volleyball

Golf Skuba diving

Hill or mountain

climbing

Basketball

Jogging Fishing

Rowing Riding

Running or athletics Movement to music

Sailing Weight training

Squash or rackets Dancing

Swimming

Spare-time activities

Chess, bridge or similar

games

Local government, trade

union or political work

Church or religious

activities

Playing a musical

instrument with others

Going to the cinema,

theatre or concerts

Voluntary social welfare

work

Helping to run a club,

playgroup or school

Table 2

Physical problems enquired about at 36 years

Bronchitis Heart trouble

Sciatica, lumbago or

recurring backache

Trouble with varicose

veins

Arthritis or rheumatism Diabetes

Persistent skin trouble

(e.g. exzema)

Trouble with gums or

mouth

Asthma Cataracts

Hay fever Stroke

Recurring stomach

trouble (e.g. ulcers)

Epilepsy

Gall bladder trouble Kidney or bladder

infections

Hernia Dizziness and

unsteadyness

Headaches or migrane Cancer

High blood pressure

M. Richards et al. / Social Science & Medicine 56 (2003) 785–792 787

other heart trouble, blood pressure problems, stroke,

transient ischaemic attack or diabetes. In addition,

the following physiological measures were obtained:

resting pulse, systolic and diastolic blood pressure,

forced expiratory volume (FEV) and forced vital

capacity (FVC).

Statistics

Multiple regression models were used to test the

association between the two kinds of activity measured

at 36 years (coded as dichotomous variables), and verbal

memory at 43 and 53 years (retained as continuous

variables). First, the association between activity and

memory at 43 years was investigated, progressively

adjusting for sex, educational attainment and social

class. To assess whether any association between activity

and memory was simply due to those of high intellectual

ability being more likely to engage in activities, the

models were further adjusted for intellectual ability at 15

years (retained as a continuous score). Formal tests

confirmed a linear relationship between this ability score

and the memory score. Finally, the models were adjusted

for health status at 36 years (physical complaints and

PSE caseness). The potential confounding effects of

employment status, marital status, number of children

living in the home, body mass index, resting pulse rate,

blood pressure or PEFR were investigated by adding

and removing these variables in turn to the models,

already adjusted for all the above background variables.

Both kinds of activity were then included in the same

model, to assess whether their associations with memory

were independent of each other.

Next, to test whether activity was associated with rate

of change in memory from 43 to 53 years, the analyses

were repeated using the memory score at 53 years as an

outcome, controlling for memory at 43 years (condi-

tional model for change), as well as sex, education,

social class, IQ at 15 years and health status at 36 and 53

years.

Finally, to test whether change in activity engagement

from 36 to 43 years was associated with rate of change in

memory from 43 to 53 years, physical exercise at 43

years was added to the above conditional model for

change for this type of activity. A comparable analysis

for spare-time activity was not possible, because the

relevant measures at 36 and 43 years were too dissimilar.

Results

Sample size and missing data

Of the 3035 cohort members interviewed at 53 years,

1116 had missing data for at least one of the variables in

the analysis. Analyses were therefore performed on the

remaining 1919 cohort members. Those with missing

data had a significantly lower general intellectual ability

at 15 years than those with complete memory data

(t ¼ 8:10; po0:001).

Activity at 36 years and memory at 43 years

Of the 1919 participants in the analysis, 36.6% had

not undertaken any physical exercise during the past

month, and 24.2% had not engaged in any of the spare-

time activities.

Unadjusted regression coefficients representing the

mean difference in the memory score at 43 years between

those engaging in activities at 36 years and those not, are

shown in Table 3.

Engagement in physical exercise and spare-time

activity at 36 years was strongly associated with higher

memory score at 43 years. These associations were not

significantly reduced by adjustment for sex (Table 3).

Indeed, that for physical exercise was strengthened, since

women engaged in this less frequently than men, but

achieved higher memory scores. Controlling for educa-

tion reduced the strength of these associations by

B50%, although they both remained significant at the

5% level. Adjusting for social class had little additional

effect. Adding baseline IQ reduced the associations for

activity further, although again, the associations re-

mained significant at the 5% level for both kinds of

activity. Final adjustment for health status (physical

disorders and significant mental distress according to the

Table 3

Estimates (regression coefficients and 95% confidence intervals)

of the effect of spare-time activity on verbal memory,

progressively adjusting for sex, education, social class, cogni-

tion at 15 years, and health statusa at 36 years

Regression

coefficient

(95% CI)

p-value

Physical exercise (baseline=none)

Unadjusted 1.57 (0.99, 2.14) o0.0001

+Sex 1.79 (1.22, 2.35) o0.0001

+Education 0.85 (0.34, 1.36) 0.001

+Social class 0.80 (0.29, 1.30) 0.002

+Cognition at 15 years 0.60 (0.12, 1.08) 0.02

+Health status 0.58 (0.10, 1.06) 0.02

Spare-time activity (baseline=none)

Unadjusted 3.22 (2.59, 3.85) o0.0001

+Sex 3.18 (2.56, 3.81) o0.0001

+Education 1.79 (1.21, 2.36) o0.0001

+Social class 1.65 (1.08, 2.22) o0.0001

+Cognition 1.39 (0.85, 1.94) o0.0001

+Health status 1.38 (0.84, 1.93) o0.0001

aPersistent physical complaints and PSE caseness.

M. Richards et al. / Social Science & Medicine 56 (2003) 785–792788

PSE) had little additional effect. Thus for a given

baseline IQ score (at 15 years), and after controlling for

sex, education, social class and health status, those who

engaged in physical exercise at 36 years scored on

average half a point more on the memory test at 43 years

than those who did no physical exercise. Similarly, those

who engaged in spare-time activity scored on average 1.4

points more.

The effect sizes for both kinds of activity were not

significantly altered by additional adjustment, in turn,

for employment status, marital status, number of

children living in the home, body mass index, resting

pulse rate, blood pressure or PEFR at 36 years.

When physical exercise and spare-time activity were

simultaneously entered into the model, along with all the

above background variables, the association between

spare-time activity and verbal memory remained sig-

nificant at the 5% level, with only a slight attenuation of

the effect, although this was no longer the case for

physical exercise (Table 3). The full model, with both

kinds of activity and the principal background variables,

is shown in Table 4.

The interaction term between physical exercise and

spare-time activity was not significant (p ¼ 0:32).To check the extent to which the effect of spare-time

activity was driven by an intellectual component, the

analysis for spare-time activity was re-run after omitting

the challenging games item (chess, bridge or similar

games) from the score. The strength of association was

slightly reduced, but remained highly significant at the

5% level (regression coefficient=1.07, 95% confidence

interval=0.54, 1.60, po0:001). However, post hoc

analysis revealed that the individual spare-time activity

most strongly associated with memory was that con-

cerned with going to the cinema, theatre or concerts (see

Table 1). Indeed, this was the most frequently endorsed

spare-time item (52.4%, compared to 25.7% for the next

most frequent category, engaging in church or religious

activities). When this item was removed from the score

the association with memory was reduced by over 50%

(regression coefficient=0.56, 95% confidence inter-

val=0.09, 1.03, p ¼ 0:02).Table 4 also shows that sex was strongly and

independently associated with memory at 43 years, with

women scoring an average of 2.4 points higher than

men. Baseline IQ, educational attainment, and social

class at age 36 were also independently associated with

memory at 43, although the effect was weakest for social

class.

Activity at 36 years and change in memory between 43 and

53 years

To assess the association between activity at 36 years

and rate of change in memory between 43 and 53 years,

the full model was repeated using memory at 53 years as

the outcome, adjusted for memory at 43 years. Results

are shown in Table 5.

Physical exercise was significantly associated with

slower rate of decline in memory score over the 10-year

period, after adjusting for spare-time activity and the

principal background variables, with those engaging in

physical activity showing an average half-point slower

decrease over those not engaging. This association was

not significantly altered by further adjusting, in turn, for

cardiovascular disease (see Methodology), resting pulse,

systolic and diastolic blood pressure, FEV and FVC at

53 years. Spare-time activity, on the other hand, was not

associated with change in memory over this interval.

To assess whether change in physical exercise engage-

ment from 36 to 43 years was associated with change in

memory from 43 to 53 years, the above analysis was

repeated, adding the similar measure of physical exercise

at 43 years. This strongly reduced the association

between physical exercise at 36 years and memory

change (regression coefficient=0.25, 95% confidence

interval=�0.18, 0.69, p ¼ 0:27). However, there was asignificant association between physical exercise at 43

years and memory change (regression coefficient=0.75,

95% confidence interval=0.33, 1.18, p ¼ 0:0005). Over-all, this suggests little protection in memory in those

who stopped physical exercise after 36 years, but

protection in those who began it after this time. Since

the interaction term between physical exercise at 36 and

43 years was not significant at the 5% level (p ¼ 0:88),

Table 4

Estimates (95% confidence intervals) from the fully adjusted

regression model of the effect of spare-time activity on verbal

memory at 43 years

B (95% CI) P

Any physical exercise 0.44 (-0.04, 0.93) 0.07

Any spare-time activity 1.32 (0.77, 1.87) o0.0001

Female sex 2.43 (1.95, 2.91) o0.0001

Educationa

0 (Baseline)

1 �0.21 (�1.10, 0.69) 0.65

2 1.37 (0.66, 2.07) 0.0002

3 1.95 (1.22, 2.68) o0.0001

4 3.65 (2.59, 4.71) o0.0001

Manual social class �0.84 (�1.39, �0.29) 0.003

Cognition at 15 0.21b (0.18, 0.24) o0.0001

Physical disorder at 36

0 (Baseline)

1 disorder �0.15 (�0.70, 0.41) 0.60

2+disorders �0.22 (�0.78, 0.34) 0.44

Mental disorderc at 36 �0.51 (�1.51, 0.49) 0.32

a0=no qualification (baseline), 1=vocational only, 2=or-

dinary (‘O’) level, 3=advanced (‘A’) level, 4=degree or

equivalent.bPer item increase in score.cPSE caseness.

M. Richards et al. / Social Science & Medicine 56 (2003) 785–792 789

the effect on memory of engaging in physical activity on

both occasions (36 and 43 years) was additive, with

those engaged at both occasions having an average

decline 1.01 points slower than those engaged at neither

age.

The association between change in spare-time activity

and change in memory was not investigated because the

measures of spare-time activity at 36 and 43 years were

not comparable.

Discussion

In a large population-based birth cohort study we

investigated associations between two kinds of activ-

ity—physical exercise and spare-time—at 36 years, and

verbal memory. Both kinds of activities were signifi-

cantly associated with memory at 43 years, and while

these associations were reduced by adjusting for sex,

educational and occupational attainment, and health

status, they remained significant at the 5% level.

However, the effect was stronger for spare-time activity

than for physical exercise, and once spare-time activity

was taken into account, there was no independent effect

of physical exercise on memory at this time. Crucially,

this effect of spare-time activity was also independent of

baseline IQ, as measured by general cognitive ability at

15 years. Thus the effect of activity at 36 years on

memory at 43 years can be interpreted as the difference

in mean memory score for a given IQ at 15 years, and

was therefore unlikely to reflect reverse causality, i.e.

those of high lifetime ability being more likely to engage

in spare-time activity.

This suggestion is reinforced by the additional finding

that engagement in physical exercise at 36 years was

associated with a slower rate of decline in memory score

between 43 and 53 years, again independently of all the

principal background variables. The effect of the 43 year

measure was stronger, so that those who gave up

exercise after 36 years did not show the same benefit as

those who exercised at 43 years (either new or

continuing). Not only does this strengthen the sugges-

tion of a causal link between physical activity and

protection of memory in midlife, but it also implies that

the cognitive benefit of physical exercise is enhanced by

persistent or more recent activity. Conversely, and

consistent with the ‘disuse’ concept of Salthouse (1991)

(more loosely, the ‘use it or lose it’ adage), these findings

suggest that this benefit is lost if activity is not

maintained. Unfortunately a comparative analysis of

change in spare-time activity would not have been

informative, since a key item asked at 36 years (going to

the cinema, theatre or concerts) was not repeated at 43

years. Nevertheless, the association between spare-time

activity at 36 years and memory at 43 years, but not

change in memory from 43 to 53 years, suggests that the

cognitive benefit from this form of engagement, once

established, remains relatively unchanged across the

ages studied.

We should highlight the disproportionate dropout of

survey members with low cognitive scores. This is a

hazard of longitudinal cognitive studies, limiting gen-

eralisability and probably leading to under-estimation of

the true population effect sizes. With this limitation in

mind, what is the nature of the association between

activity and memory?

To begin with, it is worth noting that neuronal

plasticity and development is by no means confined to

early life. For example, licensed London taxi drivers,

who are required to undertake intensive navigational

study of the city as part of their training, show

significantly larger posterior hippocampi than controls,

the size correlating with amount of occupational

experience (Maguire et al., 2000). Benefits thus accrued

in adult life may increase cognitive reserve (Stern,

Alexander, Prohovnik, & Mayeux, 1992), conferring

protection against cognitive decline. The paths through

which different kinds of activity influence cognition may

be diverse, however.

One possible way in which physical exercise protects

cognitive function is through increased cerebral oxyge-

nation, leading to improved neurotransmitter metabo-

lism (Dustman, Emmerson, & Shearer, 1990). We did

not find that measures of lung function modified the

association between physical exercise and memory in the

present study, either at 36 or 53 years. The possibility of

Table 5

Estimates (95% confidence intervals) from the fully adjusted

regression model of the effect of spare-time activity on change

in verbal memory from 43 to 53 years

B (95% CI) p

Any physical exercise 0.44 (0.01, 0.87) 0.04

Any spare-time activity 0.06 (�0.43, 0.55) 0.82

Female sex 1.44 (1.00, 1.87) o0.0001

Educationa 0 (Baseline)

1 �0.30 (�1.10, 0.49) 0.46

2 0.97 (0.34, 1.60) 0.002

3 0.91 (0.25, 1.56) 0.007

4 0.92 (�0.04, 1.87) 0.06

Manual social class �0.85 (�1.34, -0.36) 0.0007

Cognition at 15 0.14b (0.11, 0.16) o0.0001

Physical disorder at 36

0 (Baseline)

1 Disorder �0.30 (�0.79, 0.19) 0.23

2+Disorders �0.21 (�0.71, 0.29) 0.41

Mental disorderc at 36 �0.40 (�1.29, 0.48) 0.37

a0=no qualification (baseline), 1=vocational only, 2=or-

dinary (‘O’) level, 3=advanced (‘A’) level, 4=degree or

equivalent.bper item increase in score.cPSE caseness.

M. Richards et al. / Social Science & Medicine 56 (2003) 785–792790

neural protection by exercise, however, is supported by

evidence from animal studies that learning potentiates

the effect of physical activity on neural growth in the

brain (Gomez-Pinilla, So, & Kesslak, 1998). An alter-

native is that exercise reduces the risk of diseases that

impair cognitive function, such as hypertension diabetes

and cardiovascular disease (Anstey & Christensen,

2000). This seems unlikely to be a crucial factor in this

relatively young cohort, however, and, indeed, the

association between physical exercise and change in

memory was not modified by controlling for a range of

cardiovascular risk factors.

An equally difficult task is to determine which aspects

of spare-time activity are important for protecting

cognition. Evidence from the Victoria Longitudinal

Study suggests that the intellectual component carries

the greatest weight (Hultsch et al., 1999), although in

our study the effect of spare-time activity was still

significant after removing an analogous item on

intellectual stimulation (chess, bridge or similar games)

from the score. Social engagement was found to protect

against cognitive decline in the New Haven EPESE

cohort (Bassuk et al., 1999). However, the social

measures in that study included exercise and working

at a hobby, which would have been classified in our

study as physical exercise and spare-time activity

(respectively). In fact most of these activities, whether

physical or non-physical, involve social interaction.

Furthermore, all involve some degree of volition. The

finding that physical exercise but not spare-time activity

protected memory minimises the role of such non-

specific components, however.

What are the public health implications of the present

results? Our study suggests that uptake of physical

exercise in young to middle adulthood benefits memory,

an aspect of cognitive function likely to be important for

conduct of activities of daily living during ageing.

Furthermore, sustained physical activity appears to

reinforce this benefit, whereas abandonment of this

activity appears to result in its loss. It is therefore

important to investigate whether uptake of physical

activity in later life can result in cognitive benefit, or

whether this is more likely to be observed following

long-established patterns of activity. Continuing follow-

up of this cohort will help to resolve this question.

Funding for the NSHD is provided by the Medical

Research Council. Data collection at 53 years was

carried out by the National Centre for Social Research.

References

Albert, M. S., Jones, K., Savage, C. R., Berkman, L., Seeman,

T., Blazer, D., & Rowe, J. W. (1995). Predictors of cognitive

change in older persons: Macarthur studies of successful

aging. Psychology and Aging, 10, 578–589.

American Psychiatric Association. (1994). Diagnostic and

statistical manual of mental disorders (4th ed.). Washington,

DC: American Psychiatric Association.

Anstey, K., & Christensen, H. (2000). Education, activity,

health, blood pressure and apolipoprotein E as predictors of

cognitive change in old age: A review. Gerontology, 46,

163–177.

Arbuckle, T. Y., Gold, D. P., Andres, D., Schwartzman, A. E.,

& Chaikelson, J. (1992). The role of psychosocial context,

age, and intelligence in memory performance of older men.

Psychology and Aging, 7, 25–36.

Bassuk, S. S, Glass, T. A., & Berkman, L. F. (1999). Social

disengagement and incident cognitive decline in community-

dwelling elderly persons. Annals of Internal Medicine, 131,

165–173.

Birren, J. E., Kinney, D. K., Schaie, K. W., & Woodruff, D. S.

(1981). Measurement of intelligence over the lifespan. In

Developmental psychology. A life-span approach (pp. 522–

534). Boston: Houghton Mifflin Company.

Carmelli, D., Swan, G. E., LaRue, A., & Eslinger, P. J. (1997).

Correlates of change in cognitive function in survivors from

the Western Collaborative Group Study. Neuroepidemiol-

ogy, 16, 285–295.

Deary, I. J., Whalley, L. J., Lemmon, H., Crawford, J. R.,

& Starr, J. M. (2000). The stability of individual differences

in mental ability from childhood to old age: Follow-up

of the 1932 Scottish mental survey. Intelligence, 28,

49–55.

Dustman, R. E., Emmerson, R. Y., & Shearer, D. E. (1990).

Aerobic fitness may contribute to CND health: Electro-

physiological, visual and neurocognitive evidence. Journal

of Neurorehabilitation, 4, 241–254.

Gold, D. P., Andres, D., Etezadi, J., Arbunckle, T. Y.,

Schwartzman, A. E., & Chaikelson, J. (1995). Structural

equation model of intellectual change and continuity and

predictors of intelligence in older men. Psychology and

Aging, 10, 294–303.

Gomez-Pinilla, F., So, V., & Kesslak, J. P. (1998).

Spatial learning and physical activity contribute to the

induction of fibroblast growth factor: Neural substrates for

increased cognition associated with exercise. Neuroscience,

85, 53–61.

Hultsch, D. F., Hertzog, C., Small, B. J., & Dixon, R. A.

(1999). Use it or lose it: Engaged lifestyle as a buffer of

cognitive decline in aging. Psychology and Aging, 14,

245–263.

Jacobs, D. M., Sano, M., Dooneief, G., Marder, K., Bell, K. L.,

& Stern, Y. (1995). Neuropsychological detection and

characterization of preclinical Alzheimer’s disease. Neurol-

ogy, 45, 957–962.

Kramer, A. F., Hahn, S., Cohen, N. J., Banich, M. T.,

McAuley, E., & Harrison, C. R., et al. (1999). Ageing,

fitness and neurocognitiove function. Nature, 400,

418–419.

Kuh, D. J. L., & Cooper, C. (1992). Physical activity at 36

years: Patterns and chilhood predictors in a longitudinal

study. Journal of Epidemiology and Community Health, 46,

114–119.

Maguire, E. A., Gadian, D. G., Johnsrude, I. S., Good, C. D.,

Ashburner, J., Frackowiak, R. S., & Frith, C. D. (2000).

Navigation-related structural change in the hippocampi of

M. Richards et al. / Social Science & Medicine 56 (2003) 785–792 791

taxi drivers. Proceedings of the National Academy of Science

USA, 97, 4398–4403.

Masur, D. M., Sliwinski, M., Lipton, R. B., Blau, A. D., &

Crystal, H. A. (1994). Neuropsychological prediction of

dementia and the absence of dementia in healthy elderly

persons. Neurology, 44, 1427–1432.

Pidgeon, D. A. (1968). Details of the fifteen years tests. In J. W.

B. Douglas, J. M. Ross, & H. R. Simpson (Eds.), All our

future (Appendix 1). London: Davies.

Pless, I. B., Crips, H. A., Davies, J. M. C., & Wadsworth, M. E.

J. (1989). Chronic physical illness in childhood and

psychological and social circumstances in adolescence and

early adult life. Developmental Medicine and Child Neurolol-

ogy, 31, 746–755.

Ritchie, K., & Touchon, J. (2000). Mild cognitive impairment:

A critical appraisal of its conceptual basis and current

nosological status. Lancet, 355, 225–228.

Salthouse, T. A. (1991). Theoretical perspectives on cognitive

aging. Hillsdale, NJ: Erlbaum.

Schaie, K. W. (1984). Midlife influences upon intellectual

functioning in old age. International Journal of Behavioral

Development, 7, 463–478.

Schaie, K. W., & Willis, S. L. (1986). Can decline in adult

intellectual functioning be reversed? Developmental Psychol-

ogy, 22, 223–232.

Stern, Y., Alexander, G. E., Prohovnik, I., & Mayeux, R.

(1992). Inverse relationship between education and parieto-

temporal perfusion deficit in Alzheimer’s disease. Annals of

Neurology, 32, 371–375.

Taylor, H. L., Jacobs, D. K., Schucker, B., Knudsen, J., Leon,

A. S., & Debacker, G. (1978). A questionnaire for the

assessment of leisure time physical activities. Journal of

Chronic Disease, 31, 741–755.

Wadsworth, M. E. J. (1991). The imprint of time: Childhood,

history and adult life. Oxford: Clarendon Press.

Wadsworth, M. E. J., Mann, S. L., Rodgers, B., Kuh, D. L.,

Hilder, W. S., & Yusuf, E. J. (1992). Loss and representa-

tiveness in a 43 year follow-up of a national birth cohort.

Journal of Epidemiology and Community Health, 46, 300–304.

Wing, J. K., Cooper, J. E., & Sartorius, N. (1974). Present state

examination. London: Cambridge University Press.

Wing, J. K., Sturt, E. (1978) The pse-id-catego system:

Supplementary manual. Mimeo, Institute of Psychiatry,

London.

M. Richards et al. / Social Science & Medicine 56 (2003) 785–792792